Founder Securities released a research report saying that the proportion of AI computing power has entered a new stage of "20% +", and with reference to the smartphone-mobile app penetration curve, AI application development is expected to be the steepest, according to Zhitong Financial APP. Reviewing the penetration rates of smartphones, mobile Internet and smart electric vehicles, from the end of 2011 to mid-2014, the penetration rate of domestic smartphones increased from 23% to 88%, followed by the outbreak of applications, and the penetration rate of domestic mobile games increased rapidly from 22% of 2012Q4 to 81% of 2014Q2. At present, the domestic AI computing power accounts for about 22.8% of the total computing power. CGAR is about 50% in 2021-2026, with reference to the popularization process of mobile Internet.The penetration rate of AI applications may exceed 80% for a long time, and the technology is more closely integrated.
Related targets: iFLYTEK (002230.SZ), flush (300033.SZ), etc.
The main points of ▍ founder Securities are as follows:
Point 1: the increase in AI application penetration has reached an inflection point, and the ceiling is expected to exceed 80%.
The proportion of AI computing power has entered a new stage of "20% +". Referring to the smartphone-mobile app penetration curve, the development of AI applications is expected to usher in the steepest one.
Permeability is an important index to measure the development stage of new technology, and the inflection point of different stages reflects the medium-term industry growth rate. AI application penetration is difficult to measure, the market generally adopts the caliber of consulting companies, and intelligent computing penetration is used to describe the downstream model and application development rhythm, so as to provide forward-looking guidance for AI industry investment as far as possible.
Reviewing the penetration rates of smartphones, mobile Internet and smart electric vehicles, from the end of 2011 to mid-2014, the penetration rate of domestic smartphones increased from 23% to 88%, followed by the outbreak of applications, and the penetration rate of domestic mobile games increased rapidly from 22% of 2012Q4 to 81% of 2014Q2. At present, the domestic AI computing power accounts for about 22.8% of the total computing power. CGAR is about 50% from 2021 to 2026. With reference to the popularization process of mobile Internet, the penetration rate of AI applications may exceed 80% for a long time, and the technology is more closely integrated.
Point 2: AI+ industry, the general trend, asynchronous parallel.
Due to the different digital foundation, model precision requirements and business characteristics, there are differences in promoting intelligent landing in different industries. The expenditure on digital transformation of Chinese enterprises was 18.89 per cent of CAGR from 2022 to 2026. At present, there are differences in the level of informatization and fault tolerance of artificial intelligence models in different industries, which leads to differences in the rhythm of intelligent promotion among industries.
Argument 3: the market space is large, but the supply is still blank. Robot, Internet, manufacturing and finance apply Changpo thick snow track for AI.
The Internet not only has a high demand for intelligent computing power, such as artificial intelligence, video games, data processing and privacy protection, but also provides computing support for other industries. In addition to the traditional IT requirements, the manufacturing industry also has more requirements for intelligent computing in scenarios such as design simulation, global digital twinning, industrial meta-universe, industrial intelligent decision-making and so on. The demand for improving quality and efficiency in the financial industry is strong, the integration of AI and financial business is further deepened, the supply of applications is constantly enriched, and the localization of software and hardware, including computing power, is promoted in an orderly manner.
Argument 4: high-quality supply to solve the pain point of demand, limited space but fast landing, AI office, intelligent education, inclusive health care and other logic is clear.
Microsoft Corp launched as part of the free update of Windows11 on September 26th, introducing Copilot into computer-side systems and applications. The Copilot enterprise version costs an extra $30 per user per month, up 83 per cent from the previous price. The digital intelligence of the education industry is conducive to the universal benefit of educational resources, and the demand for intelligent computing is focused on intelligent campus, scientific research computing and artificial intelligence.
In the next stage of the information construction of medical institutions, the demand for computing power is increasing step by step, which is reflected in the upgrading of the original hospital-end computing hardware and more reliance on cloud computing. The medical application of B-terminal AI+ focuses on research and development and auxiliary diagnosis and treatment, and the demand for intelligent computing power is also significant in the medium and long term.
Risk Tips:
The development of artificial intelligence technology is not as expected, the speed of enterprise digital transformation is not as expected, the competition in the computing service industry aggravates the risk, the policy support is not as strong as expected, and the market expansion of cloud computing manufacturers is not as expected.